is compiling your own kernel useless or smart

On 6 Sep 2006 13:17:00 -0700, gavino staggered into the Black Sun and said:

It's so poorly documented that there's been a Kernel-HOWTO that explains the whole process since at least 1998. ... well, there *was* one, but compile a kernel" coughs up a lot of useful-looking pages.

I compiled a kernel when I'd been using Linux for less than 1 month. I got something that worked just fine after 5 tries. I'm probably better at doing this sort of thing than you are, but I wasn't as experienced in mid-1999. (And if you follow the directions on the more recent pages found by the above google, you'll avoid most of the stupid mistakes that I made, like trying to make ELF binary support a module.)

The performance differences between a distro-supplied kernel and a kernel that you've built yourself are pretty minimal in most cases. The big wins usually come from getting rid of the initrd, short-circuiting kudzu or whatever "hardware detection process" runs on boot, and getting rid of the distro-patched kernel for a vanilla kernel. Seriously, distro-patched kernels tend to have weird bugs that are just not present in the vanilla sources, just because the vanilla sources get *a lot* more testing. HTH,

-- Matt GThere is no Darkness in Eternity-But only Light too dim for us to see Brainbench MVP for Linux Admin mail: TRAP + SPAN don't belong ----------------------------- not the kind of person I'm preaching to.